tts-generation
Use MLX Audio TTS workflows for generation, voice cloning, streaming, batching, and model-specific controls.
- License 许可
- MIT
- 最近更新
- 2026-10-08
- Stars
- ★ 11
- Forks
- 3
01 前言 Front Matter
- name
- tts-generation
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Use MLX Audio TTS workflows for generation, voice cloning, streaming, batching, and model-specific controls.
- disable-model-invocation
- true
02 资源文件 3
- 参考文档references/api-reference.md1.7 KB
- 参考文档references/troubleshooting.md1.5 KB
- 参考文档references/workflows.md1.9 KB
03 SKILL.md 正文
TTS Generation
Use this sub-skill when the user wants to synthesize speech from text, clone a voice from a reference clip, stream audio during generation, or choose the right TTS model family and flags.
Route Here For
- Basic text-to-speech generation from CLI or Python.
- Voice cloning with
ref_audioand optionalref_text. - Streaming output, chunk joining, save behavior, and playback planning.
- Model-family questions for Kokoro, Qwen3-TTS, OmniVoice, Higgs Audio, CSM / MisoTTS, Spark, Ming Omni, Voxtral TTS, Chatterbox, and other TTS models shipped by the package.
- Safe command planning for
mlx_audio.tts.generate.
Route Elsewhere
- For transcription, forced alignment, streaming ASR, or WER, use
../stt-transcription/. - For enhancement, separation, VAD, or audio I/O, use
../speech-transforms-vad/. - For server endpoints or conversion, use
../server-and-conversion/.
Fast Paths
- See
references/workflows.mdfor the main TTS recipes. - See
references/api-reference.mdfor stable kwargs and model-loading facts. - See
references/troubleshooting.mdfor import, playback, and cloning failures. - Use
scripts/tts_command_builder.pyto shape a safe CLI command before a long run.
Default Safety Policy
Prefer a small reproducible recipe first: confirm the model id, confirm the reference clip path, confirm whether the model requires ref_text, and then decide whether streaming or file saving is actually needed. If the task is only about a command plan, do not start an inference run.
04 同仓库更多 Skills
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
python-api
Programmatic Object-Detection-Metrics API guidance for BoundingBox, BoundingBoxes, Evaluator, enums, coordinate conversion, AP/mAP semantics, and source-style import diagnostics.
file-evaluation
Folder-based PASCAL VOC AP/mAP evaluation from ground-truth and detection text files.
object-detection-metrics
Guides legacy Object-Detection-Metrics PASCAL VOC AP/mAP evaluation from detection text folders and direct Python API classes.
nx-workspace-patterns
Configure and optimize Nx monorepo workspaces. Use when setting up Nx, configuring project boundaries, optimizing build caching, or implementing affected commands.